Can Gemini process links to YT videos to analyze and summarize its contents. It can process uploaded videos, but I want to use the audio and video of a YT video to summarize academic / ted blogs, etc.
Hey there and welcome to the forum!
No, it cannot read links or process videos from links yet, as it’s not connected to the internet.
Keep in mind, this is for Gemini’s APIs, not any of the Gemini apps proper. I believe the mobile apps and web app have these capabilities, but there’s no way to do that in Google AI Studio or through the API without developing your own custom function calling tool.
Yes, I wanted to do that too, Johan. So far, I’ve downloaded a YouTube video as an MP4 file and uploaded it to Gemini 1.5 Pro using both AI Studio and Python code by adapting their Quickstart Google Colab notebook here. It does a good job of creating transcripts, summaries, etc. Although downloading and uploading is required, it could be automated with Python that works off the web link as you describe.
Alternatively, I developed a free open-source web app called ReadTube that might suit your needs. It processes YouTube video links and uses GPT-4o to generate a readable transcript and summary.
I implemented a similar type of approach with ytdl for nodejs. It works locally but when I deploy it in serverless youtube catches it as a bot. any suggestion on that ?